2:15 PM - 2:30 PM
[SSS06-03] Event location by waveform stacking using integrated fiber optic-seismometer networks
Keywords:Distributed Fiber Optic Sensing , Event location, Hybrid fiber-seismometer networks
Distributed Fiber-Optic Sensing (DFOS) technology provides unprecedented dense coverage of the seismic wavefield, contributing substantially to seismic monitoring. Traditional seismological workflows for event location rely on phase picks from a set of sensors and can be readily applied to the dense, individual DFOS recordings (e.g., fiber channels). However, their effectiveness may be hindered by time- and space-varying noise sources, such as optical noise, cable coupling inhomogeneities, and directional sensitivity along the cable axis, all of which result in complex body-phase arrival time statistics. Additionally, pick-based methods do not fully exploit DFOS's key advantage: the spatial coherence of signals.
Alternative approaches for event location, based on waveform stacking along predicted traveltimes, offer potential advantages for spatially dense sensors. These procedures rely on the computation of 3D traveltime lookup tables for each station. However, DFOS systems typically feature far more sensors than even the densest seismometer networks, potentially reaching the limits of available computing capacity or slowing down the location process. Addressing these challenges is fundamental to integrate DFOS measurements with conventional seismometers as part of next-generation seismic networks. One solution may involve improving the current computational resources. Alternatively, efficiency with current processing power can be achieved through dimensionality reduction of the traveltime lookup tables, supported by the usual lack of accurate 3D models.
Here, we present preliminary tests with a waveform stacking-based event locator designed to efficiently integrate DFOS data with traditional seismometers. We optimize the method's performance on data-intensive DFOS by assuming cylindrical symmetry in the velocity model, allowing for the calculation of a single 2D traveltime lookup table (horizontal distance and depth) spanning the entire domain. A 3D location grid is maintained to interpolate individual energy stack contributions from each sensor, resulting in a coherence map. To mitigate potential biases in event location due to the disparity in measurement density between DFOS and seismometers, we propose combining their contributions in the coherence matrix at the final stage of the location process, achieving a 1-to-1 balance. Alternative weighting schemes may also be explored. Testing includes assessing the noise resilience of an existing coherence-based locator that assumes a 3D model and uses a limited number of DFOS sensors, as well as experiments using real data from hybrid fiber-optic and conventional seismic networks deployed both onshore and offshore.
Alternative approaches for event location, based on waveform stacking along predicted traveltimes, offer potential advantages for spatially dense sensors. These procedures rely on the computation of 3D traveltime lookup tables for each station. However, DFOS systems typically feature far more sensors than even the densest seismometer networks, potentially reaching the limits of available computing capacity or slowing down the location process. Addressing these challenges is fundamental to integrate DFOS measurements with conventional seismometers as part of next-generation seismic networks. One solution may involve improving the current computational resources. Alternatively, efficiency with current processing power can be achieved through dimensionality reduction of the traveltime lookup tables, supported by the usual lack of accurate 3D models.
Here, we present preliminary tests with a waveform stacking-based event locator designed to efficiently integrate DFOS data with traditional seismometers. We optimize the method's performance on data-intensive DFOS by assuming cylindrical symmetry in the velocity model, allowing for the calculation of a single 2D traveltime lookup table (horizontal distance and depth) spanning the entire domain. A 3D location grid is maintained to interpolate individual energy stack contributions from each sensor, resulting in a coherence map. To mitigate potential biases in event location due to the disparity in measurement density between DFOS and seismometers, we propose combining their contributions in the coherence matrix at the final stage of the location process, achieving a 1-to-1 balance. Alternative weighting schemes may also be explored. Testing includes assessing the noise resilience of an existing coherence-based locator that assumes a 3D model and uses a limited number of DFOS sensors, as well as experiments using real data from hybrid fiber-optic and conventional seismic networks deployed both onshore and offshore.